annotate tools/protein_analysis/seq_analysis_utils.py @ 31:3c9bbbc6cbbe draft

v0.2.11 take two; removed tabs in README file
author peterjc
date Thu, 21 Sep 2017 11:23:01 -0400
parents 6d9d7cdf00fc
children 20da7f48b56f
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
1 """A few useful functions for working with FASTA files and running jobs.
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
2
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
3 This module was originally written to hold common code used in both the TMHMM
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
4 and SignalP wrappers in Galaxy.
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
5
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
6 Given Galaxy currently supports Python 2.4+ this cannot use the Python module
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
7 multiprocessing so the function run_jobs instead is a simple pool approach
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
8 using just the subprocess library.
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
9 """
30
6d9d7cdf00fc v0.2.11 Job splitting fast-fail; RXLR tools supports HMMER2 from BioConda; Capture more version information; misc internal changes
peterjc
parents: 29
diff changeset
10
6d9d7cdf00fc v0.2.11 Job splitting fast-fail; RXLR tools supports HMMER2 from BioConda; Capture more version information; misc internal changes
peterjc
parents: 29
diff changeset
11 from __future__ import print_function
6d9d7cdf00fc v0.2.11 Job splitting fast-fail; RXLR tools supports HMMER2 from BioConda; Capture more version information; misc internal changes
peterjc
parents: 29
diff changeset
12
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
13 import os
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
14 import subprocess
30
6d9d7cdf00fc v0.2.11 Job splitting fast-fail; RXLR tools supports HMMER2 from BioConda; Capture more version information; misc internal changes
peterjc
parents: 29
diff changeset
15 import sys
6d9d7cdf00fc v0.2.11 Job splitting fast-fail; RXLR tools supports HMMER2 from BioConda; Capture more version information; misc internal changes
peterjc
parents: 29
diff changeset
16
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
17 from time import sleep
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
18
29
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
19 __version__ = "0.0.2"
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
20
7
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
21 try:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
22 from multiprocessing import cpu_count
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
23 except ImportError:
29
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
24 # Must be under Python 2.5, this is copied from multiprocessing:
7
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
25 def cpu_count():
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
26 """Returns the number of CPUs in the system."""
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
27 if sys.platform == 'win32':
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
28 try:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
29 num = int(os.environ['NUMBER_OF_PROCESSORS'])
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
30 except (ValueError, KeyError):
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
31 num = 0
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
32 elif 'bsd' in sys.platform or sys.platform == 'darwin':
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
33 comm = '/sbin/sysctl -n hw.ncpu'
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
34 if sys.platform == 'darwin':
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
35 comm = '/usr' + comm
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
36 try:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
37 with os.popen(comm) as p:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
38 num = int(p.read())
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
39 except ValueError:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
40 num = 0
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
41 else:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
42 try:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
43 num = os.sysconf('SC_NPROCESSORS_ONLN')
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
44 except (ValueError, OSError, AttributeError):
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
45 num = 0
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
46
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
47 if num >= 1:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
48 return num
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
49 else:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
50 raise NotImplementedError('cannot determine number of cpus')
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
51
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
52
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
53 def thread_count(command_line_arg, default=1):
30
6d9d7cdf00fc v0.2.11 Job splitting fast-fail; RXLR tools supports HMMER2 from BioConda; Capture more version information; misc internal changes
peterjc
parents: 29
diff changeset
54 """Determine number of threads to use from the command line args."""
7
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
55 try:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
56 num = int(command_line_arg)
29
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
57 except ValueError:
7
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
58 num = default
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
59 if num < 1:
29
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
60 sys.exit("Threads argument %r is not a positive integer" % command_line_arg)
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
61 # Cap this with the pysical limit of the machine,
7
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
62 try:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
63 num = min(num, cpu_count())
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
64 except NotImplementedError:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
65 pass
29
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
66 # For debugging,
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
67 # hostname = os.environ.get("HOSTNAME", "this machine")
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
68 # sys.stderr.write("Using %i cores on %s\n" % (num, hostname))
7
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
69 return num
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
70
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
71
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
72 def fasta_iterator(filename, max_len=None, truncate=None):
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
73 """Simple FASTA parser yielding tuples of (title, sequence) strings."""
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
74 handle = open(filename)
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
75 title, seq = "", ""
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
76 for line in handle:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
77 if line.startswith(">"):
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
78 if title:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
79 if truncate:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
80 seq = seq[:truncate]
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
81 if max_len and len(seq) > max_len:
29
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
82 raise ValueError("Sequence %s is length %i, max length %i"
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
83 % (title.split()[0], len(seq), max_len))
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
84 yield title, seq
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
85 title = line[1:].rstrip()
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
86 seq = ""
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
87 elif title:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
88 seq += line.strip()
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
89 elif not line.strip() or line.startswith("#"):
29
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
90 # Ignore blank lines, and any comment lines
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
91 # between records (starting with hash).
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
92 pass
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
93 else:
25
41a42022f815 Uploaded v0.2.6, embedded citations
peterjc
parents: 7
diff changeset
94 handle.close()
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
95 raise ValueError("Bad FASTA line %r" % line)
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
96 handle.close()
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
97 if title:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
98 if truncate:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
99 seq = seq[:truncate]
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
100 if max_len and len(seq) > max_len:
29
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
101 raise ValueError("Sequence %s is length %i, max length %i"
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
102 % (title.split()[0], len(seq), max_len))
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
103 yield title, seq
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
104 raise StopIteration
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
105
29
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
106
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
107 def split_fasta(input_filename, output_filename_base, n=500, truncate=None, keep_descr=False, max_len=None):
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
108 """Split FASTA file into sub-files each of at most n sequences.
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
109
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
110 Returns a list of the filenames used (based on the input filename).
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
111 Each sequence can also be truncated (since we only need the start for
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
112 SignalP), and have its description discarded (since we don't usually
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
113 care about it and some tools don't like very long title lines).
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
114
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
115 If a max_len is given and any sequence exceeds it no temp files are
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
116 created and an exception is raised.
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
117 """
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
118 iterator = fasta_iterator(input_filename, max_len, truncate)
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
119 files = []
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
120 try:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
121 while True:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
122 records = []
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
123 for i in range(n):
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
124 try:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
125 records.append(iterator.next())
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
126 except StopIteration:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
127 break
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
128 if not records:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
129 break
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
130 new_filename = "%s.%i.tmp" % (output_filename_base, len(files))
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
131 handle = open(new_filename, "w")
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
132 if keep_descr:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
133 for title, seq in records:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
134 handle.write(">%s\n" % title)
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
135 for i in range(0, len(seq), 60):
29
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
136 handle.write(seq[i:i + 60] + "\n")
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
137 else:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
138 for title, seq in records:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
139 handle.write(">%s\n" % title.split()[0])
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
140 for i in range(0, len(seq), 60):
29
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
141 handle.write(seq[i:i + 60] + "\n")
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
142 handle.close()
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
143 files.append(new_filename)
29
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
144 # print "%i records in %s" % (len(records), new_filename)
30
6d9d7cdf00fc v0.2.11 Job splitting fast-fail; RXLR tools supports HMMER2 from BioConda; Capture more version information; misc internal changes
peterjc
parents: 29
diff changeset
145 except ValueError as err:
29
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
146 # Max length failure from parser - clean up
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
147 try:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
148 handle.close()
29
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
149 except Exception:
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
150 pass
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
151 for f in files:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
152 if os.path.isfile(f):
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
153 os.remove(f)
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
154 raise err
7
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
155 for f in files:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
156 assert os.path.isfile(f), "Missing split file %r (!??)" % f
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
157 return files
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
158
29
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
159
30
6d9d7cdf00fc v0.2.11 Job splitting fast-fail; RXLR tools supports HMMER2 from BioConda; Capture more version information; misc internal changes
peterjc
parents: 29
diff changeset
160 def run_jobs(jobs, threads, pause=10, verbose=False, fast_fail=True):
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
161 """Takes list of cmd strings, returns dict with error levels."""
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
162 pending = jobs[:]
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
163 running = []
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
164 results = {}
30
6d9d7cdf00fc v0.2.11 Job splitting fast-fail; RXLR tools supports HMMER2 from BioConda; Capture more version information; misc internal changes
peterjc
parents: 29
diff changeset
165 skipped = []
7
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
166 if threads == 1:
29
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
167 # Special case this for speed, don't need the waits
7
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
168 for cmd in jobs:
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
169 results[cmd] = subprocess.call(cmd, shell=True)
5e62aefb2918 Uploaded v0.1.2 to Test Tool Shed
peterjc
parents: 6
diff changeset
170 return results
30
6d9d7cdf00fc v0.2.11 Job splitting fast-fail; RXLR tools supports HMMER2 from BioConda; Capture more version information; misc internal changes
peterjc
parents: 29
diff changeset
171 failed = False
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
172 while pending or running:
29
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
173 # See if any have finished
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
174 for (cmd, process) in running:
29
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
175 return_code = process.poll() # non-blocking
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
176 if return_code is not None:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
177 results[cmd] = return_code
30
6d9d7cdf00fc v0.2.11 Job splitting fast-fail; RXLR tools supports HMMER2 from BioConda; Capture more version information; misc internal changes
peterjc
parents: 29
diff changeset
178 if return_code:
6d9d7cdf00fc v0.2.11 Job splitting fast-fail; RXLR tools supports HMMER2 from BioConda; Capture more version information; misc internal changes
peterjc
parents: 29
diff changeset
179 failed = True
29
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
180 running = [(cmd, process) for (cmd, process) in running
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
181 if cmd not in results]
3
fe10f448d641 Migrated tool version 0.0.6 from old tool shed archive to new tool shed repository
peterjc
parents: 0
diff changeset
182 if verbose:
30
6d9d7cdf00fc v0.2.11 Job splitting fast-fail; RXLR tools supports HMMER2 from BioConda; Capture more version information; misc internal changes
peterjc
parents: 29
diff changeset
183 print("%i jobs pending, %i running, %i completed" %
6d9d7cdf00fc v0.2.11 Job splitting fast-fail; RXLR tools supports HMMER2 from BioConda; Capture more version information; misc internal changes
peterjc
parents: 29
diff changeset
184 (len(pending), len(running), len(results)))
29
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
185 # See if we can start any new threads
30
6d9d7cdf00fc v0.2.11 Job splitting fast-fail; RXLR tools supports HMMER2 from BioConda; Capture more version information; misc internal changes
peterjc
parents: 29
diff changeset
186 if pending and failed and fast_fail:
6d9d7cdf00fc v0.2.11 Job splitting fast-fail; RXLR tools supports HMMER2 from BioConda; Capture more version information; misc internal changes
peterjc
parents: 29
diff changeset
187 # Don't start any more jobs
6d9d7cdf00fc v0.2.11 Job splitting fast-fail; RXLR tools supports HMMER2 from BioConda; Capture more version information; misc internal changes
peterjc
parents: 29
diff changeset
188 if verbose:
6d9d7cdf00fc v0.2.11 Job splitting fast-fail; RXLR tools supports HMMER2 from BioConda; Capture more version information; misc internal changes
peterjc
parents: 29
diff changeset
189 print("Failed, will not start remaining %i jobs" % len(pending))
6d9d7cdf00fc v0.2.11 Job splitting fast-fail; RXLR tools supports HMMER2 from BioConda; Capture more version information; misc internal changes
peterjc
parents: 29
diff changeset
190 skipped = pending
6d9d7cdf00fc v0.2.11 Job splitting fast-fail; RXLR tools supports HMMER2 from BioConda; Capture more version information; misc internal changes
peterjc
parents: 29
diff changeset
191 pending = []
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
192 while pending and len(running) < threads:
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
193 cmd = pending.pop(0)
3
fe10f448d641 Migrated tool version 0.0.6 from old tool shed archive to new tool shed repository
peterjc
parents: 0
diff changeset
194 if verbose:
30
6d9d7cdf00fc v0.2.11 Job splitting fast-fail; RXLR tools supports HMMER2 from BioConda; Capture more version information; misc internal changes
peterjc
parents: 29
diff changeset
195 print(cmd)
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
196 process = subprocess.Popen(cmd, shell=True)
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
197 running.append((cmd, process))
29
3cb02adf4326 v0.2.9 Python style improvements
peterjc
parents: 26
diff changeset
198 # Loop...
6
39a6e46cdda3 Migrated tool version 0.0.9 from old tool shed archive to new tool shed repository
peterjc
parents: 3
diff changeset
199 sleep(pause)
3
fe10f448d641 Migrated tool version 0.0.6 from old tool shed archive to new tool shed repository
peterjc
parents: 0
diff changeset
200 if verbose:
30
6d9d7cdf00fc v0.2.11 Job splitting fast-fail; RXLR tools supports HMMER2 from BioConda; Capture more version information; misc internal changes
peterjc
parents: 29
diff changeset
201 print("%i jobs completed" % len(results))
6d9d7cdf00fc v0.2.11 Job splitting fast-fail; RXLR tools supports HMMER2 from BioConda; Capture more version information; misc internal changes
peterjc
parents: 29
diff changeset
202 assert set(jobs) == set(results).union(skipped)
0
a2eeeaa6f75e Migrated tool version 0.0.1 from old tool shed archive to new tool shed repository
peterjc
parents:
diff changeset
203 return results